1 / 31

Universiteit Twente

Universiteit Twente. Juggling Word Graphs A method for modeling the meaning of sentences using extended knowledge graphs. Overview. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results, conclusions, future work.

Télécharger la présentation

Universiteit Twente

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Universiteit Twente Juggling Word Graphs A method for modeling the meaning of sentences using extended knowledge graphs.

  2. Overview • Introduction • Knowledge Graphs • Processing Language • Syntactic unification • Semantic evaluation • Semantic unification • Results, conclusions, future work ICCS ’02, Borovets

  3. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions University of Twente Introduction • Parlevink: • Computer Sciences • Computational Language • Dialogue systems in virtual environments • Faculty of mathematics: • Knowledge Graphs

  4. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Introduction Knowledge Graphs: • Theoretical work on models of semantics and their mathematical properties, prof. Hoede et al.

  5. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Introduction • No linguistics • No concrete applications • No automatic procedures

  6. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Introduction Project: Building a system for automatic processing of Knowledge Graphs in a NLP environment.

  7. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Intension Extension Ext() Extensional semantics Intensional semantics man Language house Knowledge Graphs Reminder the intensional triangle

  8. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Knowledge Graphs Relations: • Abstractions over human thinking • Low level: no overlap, not divisible

  9. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Knowledge Graphs Link weights describe the relevance of an aspect for: • Determining extension of a concept • Comparing concepts

  10. CHANGING COLOR equ par par - equ Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions ali ali COLOR par par ord ali ali TIME Knowledge Graphs Example: painting as “causing a change of color” ANIMATE ALI CAU

  11. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Processing Language Grammar S VP NP V Parsing Lexicon

  12. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Possible sentence graphs Processing Language “The man breaks the glass” Syntactical Unification Semantical evaluation Semantical unification Word graphs

  13. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Possible sentence graphs Syntactic Unification “The man breaks the glass” Syntactical Unification Semantical evaluation Semantical unification Word graphs

  14. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Syntactic Unification “The man paints the wall” CHANGING COLOR HUMAN equ ANIMATE ALI MALE ALI par par CAU - equ ADULT ali ali COLOR par par ord ali ali TIME

  15. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Syntactic Unification Role node: a node in a graph serving as a connection point for other word graphs with a certain grammatical function.

  16. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Syntactic Unification • Role nodes are only syntactic • Subject, object, head

  17. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Syntactic Unification • Semantical relationships stem from the place of a role node within the graph structure • No roles for agent, location, instrument

  18. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Possible sentence graphs Semantic Evaluation “The man breaks the glass” Syntactical Unification Semantical evaluation Semantical unification Word graphs

  19. “kill” “Oswald” ali LIVING ENTITY PERSON LIVING ENTITY ali par ali A SELFCONCIOUS Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Semantic Evaluation Example: “The president kills Oswald”

  20. “kill” “Oswald” LIVING ENTITY PERSON ali ali ali ali LIVING ENTITY A ali par SELFCONCIOUS Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Semantic Evaluation

  21. “kill” “Oswald” LIVING ENTITY ali ali ali A PERSON par SELFCONCIOUS Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Semantic Evaluation

  22. BREAK POSPAR EQU HUMAN ALI SUB BEVERAGE PAR MALE PAR GLASS ADULT ALI FRAGILE PAR Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions TRANSPARANT PAR FPAR ALI -PAR FRAGILE STONE LIVING ENTITY ALI 3 1 2 CAU -PAR PAR PAR FRAGILE BROKEN Semantic Evaluation

  23. HUMAN MALE ADULT ALI PAR PAR BREAK POSPAR Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions EQU SUB BEVERAGE GLASS ALI FRAGILE PAR SUB TRANSPARANT PAR LIVING ENTITY ALI -PAR FRAGILE STONE ALI 3 1 2 CAU -PAR PAR PAR FRAGILE BROKEN Semantic Evaluation

  24. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Possible sentence graphs Semantic Unification “The man breaks the glass” Syntactical Unification Semantical evaluation Semantical unification Word graphs

  25. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Semantic Unification BREAK POSPAR LIVING ENTITY EQU HUMAN ALI SUB ALI CAU BEVERAGE PAR 3 MALE 1 2 CAU PAR -PAR GLASS ADULT ALI PAR PAR FRAGILE FRAGILE PAR BROKEN TRANSPARANT PAR FPAR ALI -PAR FRAGILE STONE

  26. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Results

  27. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Results First tests: • Small lexicon & grammar • Ambiguities

  28. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Conclusions • First results are good • More testing needed • Larger lexicon & grammar

  29. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Applications • Ambiguity resolution in NLP systems • Anaphor & coreference resolution • Information Extraction

  30. Introduction Knowledge Graphs Processing Language Syntactic unification Semantic evaluation Semantic unification Results & conclusions Future Work • Building a larger lexicon • Automated lexicon learning • Testing in dialog application (Virtual Music Centre)

  31. Questions & discussion

More Related